Key Takeaways
- Alternative data is not illegal in itself, but data that is not publicly obtainable, such as credit card transaction or app usage data, carries a greater risk of being material nonpublic information.
- In the App Annie case, the SEC charged the data vendor, not its hedge fund clients, for misrepresenting that its data was aggregated and anonymized.
- For fund managers, a common exposure is Advisers Act Section 204A, which requires policies and procedures reasonably designed to prevent misuse of material nonpublic information.
Investment funds now trade on satellite images, scraped websites, app usage, and credit card data. When does that alternative data cross the line from clever research into a securities law problem?
Joel Cohen talks with Igor Rozenblit, managing partner of the regulatory consulting firm Iron Road Partners and a founder and former co-head of the private funds unit in the SEC's Division of Examinations, about how the SEC views alternative data and where fund managers face the most risk.
Is the Data Material Nonpublic Information?
Mr. Rozenblit explains that there is no government definition of alternative data; to the SEC, data is just data. Traditional sources such as 10-Qs and 10-Ks pose little risk, so examiners look at everything else. The threshold question is whether a data set is material nonpublic information, a hazy area without hard and fast rules. His rule of thumb: data obtained in a public way is safer. Satellite images that anyone could obtain are probably not MNPI, while credit card transactions that are not readily accessible run that risk. App data may sit closer to the line, and location data a bit farther away. Unless a manager is comfortable defending a data set as public, he advises, it should treat it as MNPI and take precautions.
The App Annie Case
App Annie, he explains, collected data on how people use phone apps and sold it to advertisers and hedge funds. It told clients the data was aggregated and anonymized when it was not, which made it unusually valuable for trading. The SEC charged the company under Section 10(b). Mr. Rozenblit notes that the case targeted the vendor rather than its fund clients because the vendor made the affirmative misstatements; a case against a client that recognized the data was not what it contracted for and used it anyway would be harder to bring.
Scraping and the Policies Rule
Internal data science teams raise their own questions, mostly around web scraping. Public web data is generally not MNPI, but he flags two practices worth review: clicking an "I am not a robot" CAPTCHA, which could be considered deceptive, and scraping behind logins, where terms of service may forbid it. Industry practice, he says, is to scrape past CAPTCHAs but not past logins. In his view, trading on nonpublic information is not illegal in itself absent a duty of confidentiality or deception, but deception in obtaining data could implicate Section 10(b) or Section 17(a). The provision he sees cited most in exams is Advisers Act Section 204A, which requires policies to prevent misuse of MNPI and, unlike insider trading, does not require proof of intent.
What to Know Now
The framework Mr. Rozenblit describes remains in place. App Annie, settled in 2021 for more than $10 million, is still the leading SEC case against an alternative data vendor, and the SEC's examination priorities have flagged alternative data since at least 2020. The SEC has also pressed broader insider trading theories: in SEC v. Panuwat, a jury in 2024 accepted the agency's "shadow trading" theory, and the appeal, argued in June 2026, remained pending in the Ninth Circuit as of this writing. On scraping, a federal court in Meta Platforms v. Bright Data held in 2024 that logged-off scraping of public data did not breach Meta's terms, while leaving login-protected scraping unaddressed, a line consistent with the industry practice he describes. Written policies for vetting data sources remain the most practical safeguard.
